Jul. 11, 2012

Statistics Lie



        Now,  just to put your doubts to rest, I have over 12 college hours in statistics/ decision science  which means nothing without some good sense. Some of my friends will probably say that's where I go wrong.


        I chose this subject because I was reading a news article and it was pointing out that 60% of the individuals on welfare were Anglo-Americans VS 40 % for minorities.  The point was to convince the reader that minorities do not utilize welfare and benefits more than the majority in the U.S. 


        Having worked for a research department my antennae immediately went up when I saw the numbers and heard the assumptions that were made.  First, How many individuals make up each group? Second, is this a case of using statistics to justify a conclusion.


        My students used to look at me with wide eyes when I pointed out that only 18% of the country is African-American and during the slavery era most weren't allowed to vote. If this is true then the only rational conclusion is that a whole bunch of fat old white men like me must have voted to abolish slavery.


        Statistics can and do provide enormous opportunities for us to analyze various research results.  On the other hand they can be used to deceive others while we try to justify our biased opinions.  One example a professor of mine used to use that is both humorous and typical.  He would posit that broken homes were the primary factor in crime increase.  He would then show us statistics that reflected that 40% of the incarcerated individuals in U.S. jails come from broken homes.  His solution to the problem was to force all happily married couples to get a divorce because 60% of the incarcerated must have come from happily married households.


        I chose this topic because we will be hearing all sorts of numbers, Percentages, Means, Medians, and Averages during this election year. 
We have to be careful and critical of any numbers unless we can verify them.  Then, even more importantly, we have to make sure they are saying what we thought they said or someone told us they said.


        One of the biggest errors the news agencies like to perpetuate is the use of raw numbers out of context.  If I tell you that there were 10 murders in my home town.  What does this say? It says nothing except that there were 10 murders in my hometown. Now, if you say I live in Houston with a population of 3,000,000 or Raymertown with a population of 150 we might be able to make some observations. Even worse is when they take raw numbers, averages, and percentages and mix them all up to "prove a point".


        I have actually had a professor come to my office and tell me he needed statistics to prove his findings.  Could I have done that?  Sure, as long as I didn't have to worry about someone examining the fundamentals.


        Let's all be more observant and question the numbers that are being thrown out to justify assumptions.  We may find we were forming our own opinions based upon faulty statistics.


        The "Grumpy Old Man"